Path analysis: application in an epidemiological study of echinococcosis in New Zealand

Abstract
SUMMARY: The method of path analysis is described in detail. Application of this analytical technique in the interpretation of causal relationships in complex biological systems is demonstrated using data from an epidemiological study of echinococcosis in New Zealand. The results identified the major causal pathways determining Echinococcus granulosus prevalence in dogs, based on multiple regression analysis of a linear causal model constructed from prior biological and epidemiological knowledge. Only ethnic and sheep husbandry variables had important direct effects on prevalence in the North Island of New Zealand, with some climatic factors (maximum temperature and relative humidity) and soil porosity acting indirectly through animal husbandry practices. It is suggested that path analysis, by permitting interaction between epidemiological theory and statistical analysis, provides a valuable additional tool to epidemiologists for the study of causal relationships among variables in multivariate systems.